Generative AI: what's changing for Swiss B2B

Generative AI doesn't change how your B2B company delivers its services; it changes how your future clients find you, evaluate you and compare you — well before any first contact. In Switzerland, this shift is already visible in the upstream phase of the buying journey, and it now separates companies that are present in model responses from those that are absent. The good news: what produces this presence is a well-known discipline, not a technology you buy.

Note revised on 25 May 2026. This page absorbs the article "AI in 2026: what impact for brands", which has been deindexed and redirected here.

Three moments when your clients are already looking for you in AI

The Swiss B2B decision-maker doesn't use generative AI to make a decision. They use it to prepare one, and three moments in that preparation have become established.

The first is discovery. A CFO looking for a provider for a specific project no longer just types "consulting firm Geneva" into Google. They phrase their question inside a generative environment: "Which Swiss firms would you recommend to lead the digital transformation of a fifty-person fiduciary?" The answer isn't a list of ten links; it's a reasoned recommendation that names a few players and pre-qualifies each one. The company absent from that answer doesn't exist in this buyer's discovery phase.

The second is evaluation. Once a shortlist has formed, the buyer returns to the models to dig into a specific player: "What is said about this company?" The model synthesises whatever its sources allow. If the company has little structured editorial presence and little third-party authority — business press, sector rankings, professional publications — the answer will be short, vague, sometimes inaccurate. That thin representation weighs on the decision underway.

The third is comparison: "Compare A and B for this need." The model builds an argumentative table, and the player with the stronger editorial presence is cast in the more favourable light — regardless of the actual quality of their services, which the model cannot measure. It's in this asymmetry that editorial presence work takes on its strategic weight.

Which Swiss sectors are most exposed?

The intensity of this shift varies by sector. Four deserve particular attention.

Financial and fiduciary services run on trust and reputation. The Swiss landscape counts several thousand mid-sized fiduciaries competing in cantonal or regional markets. A fiduciary absent from generated recommendations in its own client base is giving up a channel its competitors are actively building.

Consulting and professional services rely on word of mouth — now mediated by the models. An executive preparing a mandate gets reasoned recommendations within seconds. The player that publishes structured editorial content and maintains third-party authority on its subjects is favoured in this mediation, with no mechanical relationship to its size.

Precision industry and microtechnology sell into global markets, where buyers source suppliers from Berlin, Boston or Singapore by querying generative environments on technical niches. The manufacturer whose technical content is clear, structured and available in the languages of its markets stands a better chance of being cited; one whose website offers only sales brochures risks being left out.

Cloud integration and managed IT services, finally, operate in a field where buyers query the models to assess fast-changing options. A Swiss provider's presence in these answers depends on its public technical documentation, the precision of the use cases it exposes, and the multilingual coherence of its website.

Lazy content no longer pays off

Generative AI has democratised the mass production of standardised content — and that's precisely what devalues it. When an entire sector publishes the same kind of content with the same tools and the same templates, differentiation disappears; generic content becomes background noise that systems filter out in favour of substantive material. The Google doctrine published on 15 May 2026 confirmed that core ranking systems make this selection on quality, not on production method^[1].

Value therefore shifts from execution to presence strategy: defining a distinctive angle, holding a reasoned position, maintaining editorial coherence over time, and getting that material to exist with credible third parties. This value can't be outsourced to a tool. It's built — and it's built first on your own site, which needs to carry factual, dated, sourced information structured to be read by machines. That's the focus of the site built to be cited project I detail elsewhere.

Swiss multilingualism becomes a genuine asset in this context: models produce different recommendations depending on the language of the query. A company covering French, German and English to an equivalent standard holds an observable advantage over competitors who neglect one of these languages — with Italian added for anyone addressing Ticino or the neighbouring Lombardy region.

And what about data protection in all this?

The use of generative AI in B2B processes touches the FADP as soon as it involves personal data: informing clients and prospects, the legal basis for profiling processing, transparency on international transfers. These obligations apply to every Swiss company, regardless of size^[2].

The editorial presence work described in this note, however, doesn't process personal data. It concerns public content — your website, your publications, your documentation — and remains fully compatible with the Swiss legal framework. The two topics intersect at the level of governance, not execution.

Where to start, and in what order

The observation framework I recommend has four stages. A baseline diagnostic: which decision-related queries concern your company, how the leading models represent you on those queries, where the gaps are. Explicit prioritisation: which gaps merit corrective work, and which don't. Editorial and structural work on the selected gaps, carried out patiently — on your own site first, with third parties afterwards. A fresh measurement at a steady cadence, quarterly for most sectors, that turns a snapshot into a trajectory.

The instrument I've refined for this purpose, the GEO Score™, is presented in full in MCVA Cahier No. 1; how the split between ranking and citation works is detailed in AI visibility: does your SME show up in ChatGPT?. And to place this project among the others — internal tools, compliance, budget — the overall roadmap appears in What should a Swiss SME do about AI in 2026?.

This shift is nothing to dramatise as an emergency. It's an operational reality managed through regular discipline — and precisely because it demands consistency, it's what distinguishes companies that consolidate their position from those that merely react to announcements.

Key takeaways

— Your B2B buyers prepare their decisions in generative models: discovery, evaluation, comparison — whoever is absent from these answers is out of the running before the first contact. — Generic content no longer differentiates; the value lies in a distinctive angle, third-party authority and multilingual coherence. — The work is steered through measurement: baseline diagnostic, prioritisation, editorial work, quarterly re-measurement.

FAQ

Do my B2B clients really use ChatGPT to find a provider?

The integration of generative AI into the upstream phase of B2B buying — research, qualification, comparison — is observable in Swiss professional usage, without needing to attach unverifiable penetration figures to it. The most honest test is direct: phrase the questions your clients are asking themselves, and see whether you show up in the answers.

Should I publish in German if my market is French-speaking Switzerland?

If your market is strictly limited to French-speaking Switzerland, no. As soon as your client base touches German-speaking Switzerland or international buyers, yes: the models treat each language separately and produce different recommendations depending on the query language. A neglected language is market share that stays invisible in your dashboards.

Is working on my presence in AI answers compliant with the FADP?

Yes. This work concerns public content — website, publications, documentation — and doesn't process personal data. The FADP applies, however, as soon as you use AI on client data: profiling, personalisation, international transfers.

How long before I see an effect?

There's no guaranteed timeline, and I'm wary of anyone who promises one. Editorial and structural work produces its effects over time; that's why measurement happens at a steady cadence, usually quarterly, to track a trajectory rather than hope for a sudden shift.

How do the models talk about your company today? The GEO Score™ pre-audit measures, free of charge, where your company appears — and doesn't appear — in the answers from ChatGPT, Claude, Perplexity, Gemini and Mistral. A quantified answer within a few days. Request a free pre-audit

Sources

[1] Google Search Central, Optimizing your website for generative AI features on Google Search, published 15 May 2026. developers.google.com/search/docs/fundamentals/ai-optimization-guide [↩]

[2] Federal Act on Data Protection (FADP), revision of 25 September 2020, entered into force 1 September 2023. www.fedlex.admin.ch/eli/cc/2022/491/fr [↩]

Jérôme Deshaie is CEO and founder of MCVA Consulting SA, an augmented agency based in Valais. Fifteen years serving major international brands, now working directly with Swiss SMEs. Background.